StockHedge.R

if (!require(BatchGetSymbols)) install.packages('BatchGetSymbols')

## Loading required package: BatchGetSymbols

## Loading required package: rvest

## Loading required package: dplyr

## 
## Attaching package: 'dplyr'

## The following objects are masked from 'package:stats':
## 
##     filter, lag

## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

## 

library(BatchGetSymbols)
library(ggplot2)
library(reshape2)
library(plyr)

## ------------------------------------------------------------------------------

## You have loaded plyr after dplyr - this is likely to cause problems.
## If you need functions from both plyr and dplyr, please load plyr first, then dplyr:
## library(plyr); library(dplyr)

## ------------------------------------------------------------------------------

## 
## Attaching package: 'plyr'

## The following objects are masked from 'package:dplyr':
## 
##     arrange, count, desc, failwith, id, mutate, rename, summarise,
##     summarize

# Download real data from Yahoo finance
# set dates
first.date <- "2011-12-31"
last.date <- "2015-12-31"

# set tickers
tickers <- c('F', 'TSLA')

l.out <- BatchGetSymbols(tickers = tickers, 
                         first.date = first.date,
                         last.date = last.date, 
                         cache.folder = file.path(tempdir(), 'BGS_Cache') ) # cache in tempdir()

## 
## Running BatchGetSymbols for:

## 
##    tickers =F, TSLA
##    Downloading data for benchmark ticker
## ^GSPC | yahoo (1|1) | Not Cached | Saving cache
## F | yahoo (1|2) | Not Cached | Saving cache - Got 100% of valid prices | Nice!
## TSLA | yahoo (2|2) | Not Cached | Saving cache - Got 100% of valid prices | Well done!

print(l.out$df.control)

## # A tibble: 2 × 6
##   ticker src   download.status total.obs perc.benchmark.dates threshold.decision
##   <chr>  <chr> <chr>               <int>                <dbl> <chr>             
## 1 F      yahoo OK                   1005                    1 KEEP              
## 2 TSLA   yahoo OK                   1005                    1 KEEP

# Reshape to get returns
df <- reshape.wide(l.out$df.tickers)
idx <- seq(1,length(df$price.adjusted$ref.date),10)
price.adjusted <- df$price.adjusted[idx,]
ret.adjusted.prices <- data.frame(F = calc.ret(price.adjusted$F)[-1], TSLA = calc.ret(price.adjusted$TSLA)[-1])


# Correlation
cov(ret.adjusted.prices)

##                 F         TSLA
## F    0.0026060088 0.0008561211
## TSLA 0.0008561211 0.0114644436

cor(ret.adjusted.prices)

##              F      TSLA
## F    1.0000000 0.1566285
## TSLA 0.1566285 1.0000000

# Construct a new portfolio 0.8 share F + 0.2 share TSLA
portfolio = 0.86*price.adjusted$F+0.14*price.adjusted$TSLA
portfolio_return = calc.ret(portfolio, ticker="F_TSLA")
ret.adjusted.prices$F_TSLA = portfolio_return[-1]
price.adjusted$F_TSLA = portfolio
ret.adjusted.prices$ref.date = price.adjusted$ref.date[-1]

# Plotting 
df_plot_price = melt(price.adjusted, id.vars = c("ref.date"), value.name = "price")

p <- ggplot(df_plot_price, aes(x = ref.date, y = price, color = variable))
p <- p + geom_line()
p <- p + facet_wrap(~variable, scales = 'free_y') 
print(p)

## histogram of bi-weekly return
df_plot_return = melt(ret.adjusted.prices, id.vars = c("ref.date"), value.name = "return" )

mu <- ddply(df_plot_return, "variable", summarise, grp.mean=mean(return))
sdv <- ddply(df_plot_return, "variable", summarise, grp.sd=sd(return))
head(mu)

##   variable    grp.mean
## 1        F 0.004986493
## 2     TSLA 0.026289045
## 3   F_TSLA 0.009047794

head(sdv)

##   variable     grp.sd
## 1        F 0.05104908
## 2     TSLA 0.10707214
## 3   F_TSLA 0.05060105

p <- ggplot(df_plot_return, aes(x = return, color = variable, fill = variable))
p <- p + geom_histogram(alpha = 0.5, position="dodge")

print(p)

## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.